For long-haul fiber optical communication, more and more systems rely on Raman amplification to extend reach in distance and capacity. There are two types of Raman amplifiers: lumped Raman amplifier and distributed Raman amplifier. Lumped Raman amplifiers utilize a dedicated, short length of fiber to provide amplification. The instantaneous gain of the amplifier can be measured by measuring the input and output power and the pump power can be suitably adjusted to keep the gain constant. Distributed Raman amplifiers rely on the transmission fiber as the gain medium. Since the input and output of the transmission fiber are at different locations, it is not feasible to measure the instantaneous Raman gain. The distributed Raman amplifier typically operates in some level of saturation. The saturation depth is a function of the total signal power, which is a function of the span loss and the number of channels. When the channel count changes due to a fiber cut or other disruptions, the gain of the distributed amplifier can change by up 1 or 2 dB. While the change in one amplifier is small, the effect will add up along a chain of amplifiers, resulting in a large change in the channel power. The change in power can cause service outages on the surviving channels until other system control loops are able to measure and re-optimize the channel powers.
Conventionally, Raman amplifier pumps and gain are set at turn-up and are not adjusted after this point. Any variation in channel powers to changes in Raman gain (e.g., due to transients) are handled by adjusting other discrete amplifiers including erbium doped fiber amplifiers (EDFAs) over a fiber link as well as any gain equalizers in the fiber link. This control relies on distributed measurement across the fiber link and is very slow (e.g., 10s to 100s of seconds). Also, existing systems are not Raman rich, so changes in gain in a few Raman amplifiers can be tolerated by absorbing the penalty in the system margin. Existing control schemes rely on optical performance monitors (OPM) in the fiber link to detect changes in the signal power. A feedback loop can estimate the corrections required and adjust the EDFA amplifier gain/tilt and dynamic gain equalizers (DGE) that are present in the signal path. The DGE and OPM elements are only deployed in a small fraction of the nodes. The feedback loop involves communication between different nodes using an optical service channel (OSC) channel and is inherently slow. The feedback loop has settling time in the order of 10s to 100s of seconds.
As Raman amplifier deployments increase, fiber links will have more and more Raman amplifiers in-line. Thus, small changes due to transients will add up along fiber links. With the conventional feedback loops, these transients will take 10s to 100s of seconds to address which will lead to network events such as lines being declared down and mesh restored and the like. Accordingly, there is a need for Raman amplifier gain compression systems and methods based on signal power monitoring.